New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Agent loop v2: Planning & Task Management (part 2) #5077

Merged

Pwuts merged 31 commits into master from performance/planning

Aug 19, 2023

Member

Pwuts commented Jul 29, 2023 •

edited

Loading

Background

The so-called agent loop of Auto-GPT currently lacks structural planning and task management. This impedes its long-term performance and also hampers the implementation of retrieval augmentation (see #3536).

Changes & other to-do's

Part 1 - Refactoring (#4799):

Rename module agent -> agents
Check out LangChain's "Plan and execute" tools (docs, code)
Refactor Agent into BaseAgent and Agent with sensible structure

Part 2 - Planning & Task Management (#5077):

Add planner agent (WIP)
~~Add executor agent (potential solution: specialize existing Agent implenmentation into an executor agent)~~
Evaluate & tweak

Parts / building blocks

The scope and diff of this PR can be magically reduced by merging the PRs below:

Pwuts added 7 commits

July 26, 2023 23:10


          Add categories to command registry

b59f6ce


          Fix tests

1390c83


          Merge branch 'command-categories' into performance/planning

b2772b8


          Clean up prompt generation

2b81524

* Rename Performance Evaluations to Best Practices
* Move specification of response format from system prompt to Agent.construct_base_prompt
* Clean up PromptGenerator class


          Add debug logging to AIConfig autogeneration

38c6fb6


          Clarify prompting and add support for multiple thought processes to A…

031519b

…gent


          Merge branch 'master' into cleanup-prompt-generation

f06b756

netlify bot commented Jul 29, 2023 •

edited

Loading

✅ Deploy Preview for auto-gpt-docs canceled.

Name	Link
🔨 Latest commit	`fb8205e`
🔍 Latest deploy log	https://app.netlify.com/sites/auto-gpt-docs/deploys/64e0d85ebffde9000809fea1

github-actions bot added the size/l label

This was referenced Jul 21, 2023

Agent loop v2: Planning & Task Management (part 1: refactoring) #4799

Merged

Improve prompting and prompt generation infrastructure #5076

Merged

Auto-GPT-Bot added the behaviour change label

Contributor

Auto-GPT-Bot commented Jul 29, 2023

You changed AutoGPT's behaviour. The cassettes have been updated and will be merged to the submodule when this Pull Request gets merged.

Pwuts added this to the v0.4.7 Release milestone

codecov bot commented Jul 29, 2023 •

edited

Loading

Codecov Report

Patch coverage: 41.61% and project coverage change: -0.77% ⚠️

Comparison is base (4dd75ca) 51.93% compared to head (fb8205e) 51.16%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #5077      +/-   ##
==========================================
- Coverage   51.93%   51.16%   -0.77%     
==========================================
  Files         117      122       +5     
  Lines        4987     5302     +315     
  Branches      671      724      +53     
==========================================
+ Hits         2590     2713     +123     
- Misses       2199     2387     +188     
- Partials      198      202       +4

Files Changed	Coverage Δ
autogpt/agents/planning_agent.py	`0.00% <0.00%> (ø)`
autogpt/commands/__init__.py	`100.00% <ø> (ø)`
autogpt/prompts/generator.py	`95.45% <ø> (ø)`
autogpt/commands/web_selenium.py	`55.46% <27.27%> (-0.19%)`	⬇️
autogpt/agents/agent.py	`47.97% <35.93%> (-2.03%)`	⬇️
autogpt/app/main.py	`41.99% <36.00%> (-1.45%)`	⬇️
autogpt/agents/base.py	`72.38% <50.00%> (+0.63%)`	⬆️
autogpt/commands/execute_code.py	`75.92% <55.55%> (+1.61%)`	⬆️
autogpt/models/agent_actions.py	`65.15% <65.15%> (ø)`
autogpt/models/context_item.py	`74.50% <74.50%> (ø)`
... and 8 more

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

Pwuts self-assigned this

Pwuts added prompt change function: task management labels

Pwuts added 2 commits

July 29, 2023 19:02


          Merge branch 'master' into performance/planning

63c08c0


          Merge branch 'cleanup-prompt-generation' into performance/planning

f317cd7

github-actions bot added size/xl and removed size/l labels

Contributor

github-actions bot commented Jul 29, 2023

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.

Contributor

Auto-GPT-Bot commented Jul 29, 2023

You changed AutoGPT's behaviour. The cassettes have been updated and will be merged to the submodule when this Pull Request gets merged.


          WIP: PlanningAgent

e811586

Contributor

github-actions bot commented Jul 29, 2023

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.


          Merge branch 'master' into performance/planning

c1d759d

github-actions bot added size/l and removed size/xl labels


          Disable message history by default on BaseAgent

3d7bdd7

Contributor

github-actions bot commented Jul 31, 2023

This pull request has conflicts with the base branch, please resolve those so we can evaluate the pull request.

lc0rp modified the milestones: v0.4.7 Release, v0.4.8


          Merge branch 'master' into performance/planning

304ec77

Contributor

github-actions bot commented Aug 9, 2023

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.

github-actions bot removed the conflicts label

Contributor

github-actions bot commented Aug 9, 2023

Conflicts have been resolved! 🎉 A maintainer will review the pull request shortly.


          Remove browsing overlay & separate browsing from extraction code

548888a

Contributor

github-actions bot commented Aug 9, 2023

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.

Pwuts added 4 commits

August 9, 2023 19:13


          Merge branch 'master' into structured-thought-processing

ffc478d


          Fix human feedback

fdfc6c3


          Fix tests

7b146a0


          Merge branch 'structured-thought-processing' into performance/planning

f9cb030

Contributor

github-actions bot commented Aug 10, 2023

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.


          Include history in Agent prompt generation

fd1e868

Contributor

github-actions bot commented Aug 14, 2023

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.


          Code improvements in agent.py

c89bf58

Contributor

github-actions bot commented Aug 15, 2023

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.


          Add ask_user command and revise system prompt

319ad2e

Contributor

github-actions bot commented Aug 19, 2023

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.


          Merge branch 'master' into performance/planning

fb8205e

Contributor

github-actions bot commented Aug 19, 2023

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.

Pwuts marked this pull request as ready for review

August 19, 2023 14:59

Member Author

Pwuts commented Aug 19, 2023

This has been tested to considerably improve the performance of Auto-GPT on the benchmark, and it contains a number of building blocks for further improvements.

Contributor

Auto-GPT-Bot commented Aug 19, 2023

You changed AutoGPT's behaviour. The cassettes have been updated and will be merged to the submodule when this Pull Request gets merged.

Pwuts merged commit 3fe2246 into master

Pwuts deleted the performance/planning branch

August 19, 2023 15:44

This was referenced Sep 8, 2023

added custom exception handling #4279

Closed

Allow asking human assistance/feedback incontinuous mode #150

Closed

Auto-GPT Performance 📈 #5190

Closed

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

behaviour change function: task management prompt change size/xl